From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262114AbTJXJ1C (ORCPT ); Fri, 24 Oct 2003 05:27:02 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262120AbTJXJ1C (ORCPT ); Fri, 24 Oct 2003 05:27:02 -0400 Received: from hermine.idb.hist.no ([158.38.50.15]:786 "HELO hermine.idb.hist.no") by vger.kernel.org with SMTP id S262114AbTJXJ07 (ORCPT ); Fri, 24 Oct 2003 05:26:59 -0400 Date: Fri, 24 Oct 2003 11:36:35 +0200 To: Daniel Phillips Cc: Jens Axboe , Linux Kernel Subject: Re: [PATCH] ide write barrier support Message-ID: <20031024093635.GA22894@hh.idb.hist.no> References: <20031013140858.GU1107@suse.de> <200310231822.36023.phillips@arcor.de> <20031023162310.GQ6461@suse.de> <200310231920.39888.phillips@arcor.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200310231920.39888.phillips@arcor.de> User-Agent: Mutt/1.5.4i From: Helge Hafting Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Oct 23, 2003 at 07:20:39PM +0200, Daniel Phillips wrote: > These are essentially the same, they both rely on draining the downstream > queues. But if we could keep the downstream queues full, bus transfers for > post-barrier writes will overlap the media transfers for pre-barrier writes, > which would seem to be worth some extra effort. > > To keep the downstream queues full, we must submit write barriers to all the > downstream devices and not wait for completion. That is, as soon as a > barrier is issued to a given downstream device we can start passing through > post-barrier writes to it. > This approach may fail: a. Some pre-barrier writes go to all devices b. barrier is sent to all devices c. Post-barrier writes go to all devices d. drive 1 commits all its pre-barrier writes, then commits its post-barrier writes. e. drive 2 is slow and havent done the pre-barrier writes yet f. power is lost - leaving inconsistent devices. The problem is that drive 1 don't know wether drive 2 did the barrier yet. Helge Hafting